Should internal-error default to dumping core?

Should internal-error default to dumping core?

[Andrew Cagney]

Hello,

At preset an internal-error interaction looks like:

> (gdb) maint internal-error
> /home/scratch/GDB/src/gdb/maint.c:119: gdb-internal-error: internal maintenance
> An internal GDB error was detected.  This may make further
> debugging unreliable.  Continue this debugging session? (y or n) n
> 
> Create a core file containing the current state of GDB? (y or n) n
 > $

that is the default/no case is to not dump core.

At present, if GDB encounters an internal error (from say a NULL pointer 
reference) it doesn't leave any physical evidence around in the form of 
a core dump.  I'm thinking it should.  Hence, I'd like to change the 
second prompt to be one that defaults to a core dump.

Looking back through the notes it appears that the [my] rationale for 
the current behavour was somewhat arbitrary.

thoughts?
Andrew

Re: Should internal-error default to dumping core?

[Fernando Nasser]

Andrew Cagney wrote:
> 
> Hello,
> 
> At preset an internal-error interaction looks like:
> 
> > (gdb) maint internal-error
> > /home/scratch/GDB/src/gdb/maint.c:119: gdb-internal-error: internal maintenance
> > An internal GDB error was detected.  This may make further
> > debugging unreliable.  Continue this debugging session? (y or n) n
> >
> > Create a core file containing the current state of GDB? (y or n) n
>  > $
> 
> that is the default/no case is to not dump core.
> 
> At present, if GDB encounters an internal error (from say a NULL pointer
> reference) it doesn't leave any physical evidence around in the form of
> a core dump.  I'm thinking it should.  Hence, I'd like to change the
> second prompt to be one that defaults to a core dump.
> 
> Looking back through the notes it appears that the [my] rationale for
> the current behavour was somewhat arbitrary.
> 
> thoughts?
> Andrew

I think as long as the default when not running interactively is to dump
core
we are OK.

Having the two defaults being reversed (n followed by a y) would be a
little
uncomfortable.  But you could change the wording instead if you really
want
to change the interactive default (I sincerely hope people are not used
to
answer this question already ;-), so there wouldn't be much impact with
the 
change).

-- 
Fernando Nasser
Red Hat Canada Ltd.                     E-Mail:  fnasser@redhat.com
2323 Yonge Street, Suite #300
Toronto, Ontario   M4P 2C9

Re: Should internal-error default to dumping core?

[Andrew Cagney]

> 
> I think as long as the default when not running interactively is to dump
> core
> we are OK.
> 
> Having the two defaults being reversed (n followed by a y) would be a
> little
> uncomfortable.  But you could change the wording instead if you really
> want
> to change the interactive default (I sincerely hope people are not used
> to
> answer this question already  [;-)] , so there wouldn't be much impact with
> the 
> change).

Sorry you've lost me here.  query() returns false if there is no tty. 
So ``n'' ``n'' would still be the default.  I'll need to change the 
question so that a ``no'' indicates a core dump.

Andrew

Re: Should internal-error default to dumping core?

[Fernando Nasser]

Andrew Cagney wrote:
> 
> Sorry you've lost me here.  query() returns false if there is no tty.
> So ``n'' ``n'' would still be the default.  I'll need to change the
> question so that a ``no'' indicates a core dump.
> 

That was how I was suggesting it should be, so it is OK with me.

-- 
Fernando Nasser
Red Hat Canada Ltd.                     E-Mail:  fnasser@redhat.com
2323 Yonge Street, Suite #300
Toronto, Ontario   M4P 2C9

Re: Should internal-error default to dumping core?

[Andrew Cagney]

> Andrew Cagney wrote:
> 
>> 
>> Sorry you've lost me here.  query() returns false if there is no tty.
>> So ``n'' ``n'' would still be the default.  I'll need to change the
>> question so that a ``no'' indicates a core dump.
>> 
> 
> 
> That was how I was suggesting it should be, so it is OK with me.
> 
Turns out I'm 180 degrees out here.  query() returns true when no TTY.


Consequently I need to reverse first query() not the second.

	Andrew

> Hello,
> 
> At preset an internal-error interaction looks like:
> 
> (gdb) maint internal-error
> /home/scratch/GDB/src/gdb/maint.c:119: gdb-internal-error: internal maintenance
> An internal GDB error was detected.  This may make further
> debugging unreliable.  Continue this debugging session? (y or n) n
> 
> Create a core file containing the current state of GDB? (y or n) n
>> $
> 
> that is the default/no case is to not dump core.
> 
> At present, if GDB encounters an internal error (from say a NULL pointer reference) it doesn't leave any physical evidence around in the form of a core dump.  I'm thinking it should.  Hence, I'd like to change the second prompt to be one that defaults to a core dump.
> 
> Looking back through the notes it appears that the [my] rationale for the current behavour was somewhat arbitrary.
> 
> thoughts?
> Andrew